By John Rydning
Storage device options continue to broaden for end users, integrators, and storage system OEMs. For enterprise applications, hard disk drives (HDDs) are now available in several form factors, configured with parallel SCSI, Fibre Channel, Serial ATA (SATA), or Serial Attached SCSI (SAS) interfaces. An increasing number of solid-state disks (SSDs) are now also shipping for enterprise applications. DRAM-based SSDs have for many years satisfied the needs of very high-performance, latency-sensitive environments, but at a high cost. Now, with the price declines in NAND flash and advances in technology, NAND-based SSDs have a growing opportunity in data-center environments. Similar to HDDs, flash-based SSDs are offered with several interface options.
Today’s diversity of storage device options for enterprise applications contrasts sharply with the limited selection available just five years ago. Consider the changes that have taken place just with hard disk drives. In 2003, essentially two HDD form factors serviced the majority of enterprise data-center storage demands: 3.5-inch 10,000 rpm and 3.5-inch 15,000rpm HDDs. Today, in addition to 3.5-inch devices, the HDD industry also ships 2.5-inch 10,000rpm and 15,000rpm enterprise-class small-form-factor (SFF) drives. Collectively, IDC classifies these drives as performance-optimized HDDs. By 2009, 2.5-inch performance-optimized HDDs will out-ship 3.5-inch drives.
HDD vendors have done an admirable job increasing the average capacity of performance-optimized disk drives, but the growth rate is slowing. At the same time, 2.5-inch performance- optimized HDDs have less storage capacity than 3.5-inch drives due to smaller diameter disks and a lower maximum number of disks per drive. Given the migration to 2.5-inch enterprise HDD products, the resulting overall average capacity of performance-optimized drives shipping to enterprise applications has remained relatively flat over the past few years.
Meanwhile, IT managers are faced with the dilemma of how to meet capacity growth requirements exceeding 50% per year. Of course, one solution was to simply increase the number of HDDs per array, arrays per system, or number of systems per datacenter to reach the required amount of capacity. Instead, the HDD industry responded by delivering high-capacity drives that store 3× to 4× more data than traditional performance-optimized drives, and are suitable for 24×7 use in rack-mount environments. Use of these capacity-optimized drives helped IT managers meet storage growth requirements and resulted in a steady increase of the average HDD capacity shipping to data centers. Capacity-optimized drives will increasingly penetrate the data center as users learn how to better leverage these devices. In 2007, about 10.8 million capacity-optimized HDDs shipped in data-center storage solutions. Over time, IDC expects a growing percentage of terabytes will be stored on capacity-optimized disk drives (see figure, above).
Meanwhile, there has been a growing disconnect between the I/O capabilities of server/controller CPUs and HDD performance. To highlight this further, consider a few of the various techniques that are employed today to increase storage performance, such as striping data across multiple disks, using HDDs with faster spin speeds (15,000rpm versus 10,000rpm), or even short-stroking the disk drive (using only the outer tracks of the disk where sequential data rates are fastest).
Generally, SSDs have much higher I/O capabilities than hard disk drives. When leveraged properly, SSDs can accelerate an entire enterprise system by ensuring the I/O capability of the storage devices is in balance with the rest of the system. IDC believes that enterprise storage applications can benefit from the use of SSDs. In particular, storage applications requiring high I/Os-per-second (IOPS) rates to support the data architecture and performance requirements of the computer system are ideal candidates for SSD-enabled acceleration.
IT managers also face a growing challenge with power and cooling. It is not easy to predict the exact cost of electricity, but it is easy to anticipate the direction: up. A number of strategies have emerged, and will continue to emerge, to help reduce the energy associated with storage. “Green” requirements will undoubtedly set in motion greener storage solutions, including at the device level.
Taken together, the result is a challenging environment for product positioning by HDD and SSD vendors. There are myriad storage device options available to end users and integrators. Some may employ a combination of performance-optimized and capacity-optimized HDDs, while some may leverage SSDs and capacity-optimized HDDs.
Given the options available today and in the near future, the permutations of storage system configurations is almost mind-boggling. Nevertheless, within this environment, standardized solutions are needed wherever possible to achieve economies of scale necessary to help IT managers cope with rising storage costs.
Serial interface adoption
Amid the confusion with device proliferation is one certainty: Storage devices, including HDDs and SSDs, are increasingly migrating to standardized serial interfaces. Historically, storage technology or form-factor transitions in the enterprise take time, interfaces included. But in comparison to other relatively recent technology transitions in the enterprise, the migration from parallel SCSI to SAS technology has been relatively quick (see figure).
In addition to the rapid transition to SAS for performance-optimized drives, the vast majority of capacity-optimized drives shipping for enterprise applications today employ the SATA interface. When accounting for all performance and capacity-optimized HDDs that shipped in 2007, more than 65% shipped with SATA or SAS interfaces.
More on SAS
The SAS interface was introduced in 2005 with a 3Gbps data rate. The 3Gbps data rate adequately supports the majority of storage systems requirements in the market today. But requirements change, and the SAS interface is being improved to meet these needs.
The second generation of the SAS interface, dubbed SAS-2, is now being readied for product launches later this year. SAS-2 will have a 6Gbps data rate and will be backward-compatible with existing 1.5Gbps SATA and 3Gbps SAS/SATA devices and infrastructure.
The 6Gbps SAS interface not only enables faster data rates, but it also offers some new benefits and opportunities for enterprise applications, including the ability to spread I/O requests over a greater number of HDDs. It also provides a higher-performance interface for future SSD devices designed with very fast I/O and data-rate capabilities. The 6Gbps SAS interface also allows for the design of SAS solutions that could compete with storage systems currently leveraging the Fibre Channel interface
Amid the multitude of transitions taking place for enterprise-class storage devices, and the growing importance of dynamics such as power and cooling, HDD and SSD OEMs will have additional market dynamics to navigate.
The success of products that integrate the various storage device types, along with pricing and product portfolio decisions, will all factor into specific HDD and SSD device demand. Despite the uncertainty within this environment, one element is certain: Data-center storage requirements for organizations continue to grow. Given the strong demand for storage capacity, the outlook for HDDs is solid, and the industry will likely post successive years of record-breaking shipments and revenue.
John Rydning is research director for hard disk drives at IDC (www.idc.com). This article originally appeared in the SCSI Trade Association’s (STA) Serial Storage Wire newsletter (www.serialstoragewire.net), and was excerpted from the IDC reports “Worldwide Hard Disk Drive 2008-2012 Forecast: Shrugging Off Storage Technology Challengers;” “Worldwide Solid State Drive 2007-2012 Forecast and Analysis: Entering the No-Spin Zone;” and “The Real Costs to Power and Cool All of the World’s External Storage.”
The state of solid state in the enterprise
By Ashish Nadkarni
Flash drives, also known as solid-state disks (SSDs), have a bright future in the enterprise space. They promise to overcome most limitations of traditional hard drives: performance, power consumption, heat dissipation, mean time between failure, etc.
There is no doubt that flash drives will eventually replace rotational hard drives in the enterprise space, as well as the consumer market. However, if you have bought into the promise of solid-state disks and are planning to invest in it right away, hold your plans for now. The technology is seemingly mature, but still has to establish itself in the enterprise space. Only then will it become a viable replacement technology for spindle-based drives.
Before we discuss the technical pros and cons of flash, let’s examine the vendor and provider landscape for solid state disks.
It is only a matter of time before current hard disk vendors become OEM suppliers for flash drives. However, considering the amount of time and money invested in traditional hard drive technology, it is difficult for these companies to do a paradigm shift into a technology that is eventually going to render much of their current investment obsolete.
Among storage systems players, EMC is the most notable vendor that offers flash drives (OEM'd from STEC) in its disk arrays. EMC claims that flash drives can form a new “Tier 0” in storage subsystems, benefiting high-performance applications that demand lots of IOPS with fast response times. However, in creating this new tier there is bound to be a greater demand for quantification of the benefits that applications existing today on other storage tiers will see. It will be interesting to see if other pieces of the application architecture can catch up or will remain the weak link in the chain, thereby suppressing the benefits of flash drives.
Capacity, speed, reliability
Flash drives are made from non-volatile NAND flash memory. Because there are no moving parts in solid-state disks, they are very fast compared to their spinning counterparts. Magnetic hard drives suffer from seek latency, which is amplified when performing lots of random IOPS, resulting in data being fetched, written to (or a combination of both) very slowly. Furthermore, latency in a rotational medium is a function of the location of data; hence, data response times can vary significantly. Flash drives do not have such limitations and therefore perform very well in such mixed workload environments.
In published documentation, EMC claims an IOPS increase of 30 times, compared to 15,000rpm Fibre Channel drives, with response times of less than two milliseconds.
All of this performance comes at a fraction of the power consumed by magnetic hard drives. A 64GB flash drive, for example, can use 30% to 40% less energy than a 73GB 15,000rpm magnetic drive. While this may not be significant in a single-drive scenario, when several drives are replaced in a storage subsystem the savings can quickly add up.
Reduced power consumption also means reduced heat dissipation. Since SSDs do not generate heat, the array as a whole will have a lower thermal footprint and reduce air-conditioning requirements.
While they may share the basic architecture with drives found in consumer electronics, enterprise-class flash drives have better internal checks and electronics that improve their reliability. EMC, for example, claims that the drives installed in its systems have internal balancing algorithms to ensure all areas in the drive are accessed uniformly.
However, unlike their magnetic counter-parts, which have well-known failure and reliability profiles, the reliability profiles of flash drives are relatively unknown. This is not to say that the drives may fail due to component failure, but there is a chance that a bug or a defect in the solid-state design or the electronics could pose reliability issues.
Reliability aside, performance and power consumption alone make flash drives a worthwhile proposition until you look at capacity. In the magnetic hard-drive space, drive capacity has reached 1TB in the case of SATA drives. Fibre Channel and SAS drives are typically in the 500GB range. But the highest capacity for SSDs is typically about 72GB (although some vendors are shipping higher-capacity SSD drives). Given that capacity requirements are not going down, it would require a lot more slot capacity to obtain the usable yield that could be realized by using a lot fewer hard drives. As flash technology matures, capacities will go up significantly, but for now capacity is very limited.
Connectivity and cost
Today, due to the fact that the primary demand for flash drives may eventually lie in the enterprise space, connectivity options may be limited. EMC, for example, uses standard Fibre Channel interfaces for its flash drives. However, interfaces such as SAS are becoming very popular and the flash industry may choose to standardize on SAS as opposed to Fibre Channel.
Of course, high cost is the biggest drawback to SSDs. Although costs will come down eventually, for now you should be prepared to write huge checks for these products.